aIn the month of October, singers, songwriters, and music makers upload 100,000 new songs every day to streaming services like Spotify. That’s a lot of music. There is no reality, alternative or otherwise, where anyone can hear it all even a thousand times. Whether you’re into Japanese noise, Russian hardcore, Senegalese afro-house, Swedish hip-hop, or Bay Area hip-hop, the sheer range of listening options available is crippling. It’s a huge problem that data scientist Glenn MacDonald is working on. In the below excerpt from Taste Computing: Algorithms and Music Makersauthor and Tufte University anthropologist Nick Seaver explores McDonald’s unique landscape-based methodology for highlighting all the trails you didn’t know you couldn’t live without.
Reprinted with permission from Taste Computing: Algorithms and Music Makers Written by Nick Seaver, Published by University of Chicago Press. © University of Chicago 2022. All rights reserved.
world of music
“We are now at the dawn of the age of boundless connected music,” declared the data chemist from below the Space Needle. Glenn MacDonald chose his moniker himself, preferring “chemistry,” with its esoteric associations, to the now usual “data science.” His job, as he described it on stage, was to “use mathematics, writing, and computers to help people understand and discover music.”
MacDonald practiced his alchemy for the Spotify music streaming service, turning the raw materials of big data—records of listener interactions, fragments of digital audio files, whatever else he could get his hands on—into precious gold: products that might attract and retain paying customers. The mystical force of McDonald’s chemistry lies in the way in which ordinary data, if handled properly, seems to transform from subtle interactive effects into dense cultural significance.
It was 2014, and McDonald was performing at the Pop Conference, an annual gathering of music critics and academics held in the crumbling stack of a Frank Gehry-designed building in downtown Seattle. I was on the other side of the country, and I followed it online. That year, the conference’s theme was “Music and Mobility,” and MacDonald began his talk by recounting his personal musical journey, playing samples as he went. “As a kid, I discovered music by steadfastness and waiting,” he began. As a child at home, he would listen to the folk music his parents were playing on the stereo. But as he got older, his listening range expanded: the car radio played heavy metal and new wave; The Internet has revealed a world of new and mysterious species to explore. Stuck where he was, a passive observer of the music that passed by, he was finally measuring his life’s progress by his ever-expanding musical horizons. MacDonald was able to turn this passion into a career, working to help others explore what he called “the world of music,” which on-demand streaming services made more accessible than ever.
Elsewhere, MacDonald (2013) describes the world of music as if it were a landscape: “Follow any path, no matter how improbable and unchallenged it seems, and you will find a hidden valley with hundreds of bands that have lived there for years, reconstructing the world of music in Systematically and idiosyncratically changing microcosm, as in Australian hip-hop, Hungarian pop, microhouse or Viking metal”.
Travelers through the world of music will find familiarity and surprise – sounds they never imagined and songs they love. MacDonald was amazed at this new ability to hear music from all over the world, from Scotland, Australia or Malawi. “The perfect music for you might come from the other side of the planet,” he said, but that wasn’t a problem: “In music, we have the teleporter.” On-demand streaming offered a kind of musical navigation, allowing listeners to travel through the world of music instantly.
However, he suggested, repeating the catchphrase, that the size of this world might be overwhelming and difficult to navigate. “In order for this new world to be truly tangible,” MacDonald said, “we have to find ways to map that space and then build machines to take you through it on interesting paths.” The recommendation systems introduced by companies like Spotify were the machines. MacDonald’s recent work has focused on maps, or as he describes it in another talk: “a kind of thin layer of clear, inscrutable order over the writhing, rising, relentlessly expanding information monster of all the world’s music.”
Although his language may have been extraordinarily poetic, MacDonald was articulating an understanding of musical diversity that is widely shared among music recommenders: music exists in a kind of space. This space, on the one hand, is somewhat mundane—like a landscape you might walk through, encountering new things as you go. But in another sense, this space is very strange: behind the valleys and hills, there is a writhing beast rising, ever growing and linking points in space together, infinitely connected. The Music Space can look as natural as the mountains seen from the top of the Space Needle; But it can also look like a man-made topological jumble at its base. It is organic and intuitive. It’s technical and messy.
Spatial metaphors provide a dominant language for thinking about differences between music recommendation makers, as in machine learning, and between Euro-American cultures in general. Within these contexts, it is easy to imagine certain things being the same as they are collected hereWhile various other things come together there. In conversations with engineers, it is very common to find the space of music called into existence through gestures, enveloping the speakers in an imaginary environment inhabited by short pinches of air and orchestrated by hand waves. There is one type to your left and another to your right. On whiteboards and windows dotted around the office, you might find the music space displayed in two dimensions, containing a collection of dots that cluster and spread across the level.
In the music space, similar music is nearby. If you find yourself in such a space, you must be surrounded by the music that you love. To find more of them, you just need to look around and move around. In the music space, genres are like regions, playlists are like tracks, and tastes are like drift regions and archipelagos. Your new favorite song may be on the horizon.
But despite their familiarity, such spaces are strange: similarities can be found anywhere, and points that seemed far apart may suddenly become adjacent. If you ask, you will learn that all of these spatial representations are just shorthand for something more complex, for a space that has not two or three dimensions but potentially thousands of them. This is McDonald’s monster of information and space, a mathematical abstraction that extends human spatial intuition beyond its breaking point.
Such spaces, commonly called “similarity spaces,” are the symbolic terrain on which machine learning works. To classify data points or recommend items, machine learning systems typically locate them in spaces, group them into groups, measure the distances between them, and draw boundaries between them. Machine learning, as cultural theorist Adrian Mackenzie (2017, 63) has argued, “makes all differences as distances and directions of movement.” So while the music space is in one sense an informal metaphor (the landscape of musical diversity) in another sense it is a highly technical formal element (the mathematical underpinning of the algorithmic recommendation).
Spatial understanding of data transmission through technical infrastructures and everyday conversations; They are at once a form of figurative expression and a concrete arithmetic exercise. In other words, “space” here is the formal—a restricted technical concept that facilitates precision through abstraction—and what anthropologist Stefan Helmrich (2016, 468) calls informality—a less disciplined metaphor that travels alongside formal techniques. In practice, it is often difficult or impossible to separate the technical specificity from the metaphorical accompaniment. When music makers talk about space, they’re talking at once figuratively and technically.
For many critics, this “engineering rationality” (Planck 2018) of machine learning makes “culture” itself anathema: it defines traits, rationalizes feelings, and wrests cultural objects from their everyday social contexts into sterile isolation from a computational network. For example, mainstream cultural anthropology has long defined itself in the face of forms like these, which seem to lack the thickness, sensitivity, or adequacy of lived experience that we seek through ethnography. As political theorists Louise Amoore and Volha Piotukh (2015, 361) suggest, such analyzes “reduce heterogeneous forms of life and data into homogeneous spaces of computation.”
To use the terminology of geographer Henri Lefebvre (1992), spaces of similarity are clear examples of an “abstract space”—a kind of representative space in which everything can be measured and quantified, and controlled by central authorities in the service of capital. Media theorist Robert Bree (2015, 16), applying Lefebvre’s framework to music streaming, notes that people like McDonald—”data analysts, programmers, and engineers”—are primarily concerned with the abstract and visualized space of computation and measurement. Perceived space is, in Lefebvrian thought, a parasite on social and subsistence space, which Prey associates with listeners who resist and reinterpret the works of technologists. The spread of abstract space under capitalism, in this context, heralds the “destructive invasion of living by the settled” (Wilson 2013).
But to the people who work with it, the music space doesn’t feel like a sterile web, even in its most calculating state. Music Makers’ recommendations are not limited to subtle abstractions of the envisioned space. During their training, they learn to experience music space as ordinary and uninhabitable, despite its underlying strangeness. The space of music is as intuitive as a walkable landscape and as grotesque as a complex, high-dimensional geometric object. To use an often dubious distinction from cultural geography, they treat ‘space’ like ‘place’, as if the abstract, homogeneous grid were some sort of livable local environment.
Spaces of similarity are the result of many decisions; They are by no means “normal,” and people like McDonald know that the choices they make can profoundly rearrange them. However, spatial metaphor, transmitted through speech and gesture, illustration, and computation, helps make patterns in cultural data seem real. Mixing maps and territories—between fluid representations and objective topography—is fruitful for people who are both interested in creating objective knowledge and interested in calculating their subjective impact on the process. These spatial understandings alter the meaning of musical concepts such as genre or social phenomena such as taste, making them forms of agglomeration.
All products recommended by Engadget are selected by our editorial team, independently of the parent company. Some of our stories include affiliate links. If you buy something through one of these links, we may earn an affiliate commission. All prices are correct at the time of publication.